Lunar Mining and Refining

The special challenges and opportunities of turning regolith into metals, glass materials, oxygen, and more.

The Moon has a marked tendency to either have lots of something, or none at all of something. Once we are able to really dig around there is reason to hope we will find some stuff there is no sign of on the surface. For the first years, it is prudent to assume what you have to work with is only what has been proven to be there.

That list is adequate for a lot of things, but some ingenuity will be required to turn it into all we need for survival. Every way we process ore must be reinvented to suit the Moon. The chemicals most used in batch reactions to refine out certain products are entirely absent from the Moon. If water and other volatiles extracted from the poles (such as carbon dioxide and ammonia) are used in chemical processing, that reduces the amount available for a range of other essential uses - fuel, life support and ecosystem development, and as ingredients of important materials. Production sufficient to satisfy local needs will take time, and they will remain precious for much longer. If alternate practices can be developed that reduce the need to use them in refining and chemicals production, supply can go to uses where there is no substitute.

When chemicals are used on Earth to produce a specific product chemical in a batch reaction, other product(s) of the reaction are often allowed to dissipate or are thrown away because they have no value - substances like water and carbon dioxide. This would never be the case on the Moon, everything has to be collected and recycled as much as possible. If - and only if - that is done efficiently, and practically, such that only minimal amounts are lost during processing, there are many cases where it would likely be preferable to periodically restock a key reactant from Earth than to use an alternative process that is more demanding.

On the other hand, many reactions only happen above a certain temperature, or happen faster or differently above a certain temperature. A reactor that operates at very high temperatures reduces the need for reactants or allows a commonly available reactant (such as oxygen) to be used in place of an imported one (such as fluorine). Because the sun is strong and constant and heat dissipates slowly in a vacuum, high-temperature reactors are an attractive option for many things. The materials needed are expensive, but the resulting reactor lasts a long time.

As the timeline progresses, some elements become much more plentiful. Not only is water mined in large volumes from the lunar poles, but also from asteroids brought into orbit. Later it is also mined from Ceres and Callisto. Processes that rely on water, or on hydrogen purified out of water, are more attractive as time progresses. This is also true of reactions involving carbon. Carbon chemicals will be mined from asteroids, and later come in from multiple other worlds more cheaply than it can be delivered from Earth. If carbon, hydrogen, or water can be very efficiently recycled and reused, it becomes worthwhile to use it, to a limited degree in Phase 2 and then much more in Phase 3 and beyond.

Then there are techniques that produce chemicals using energy alone. In its simplest form, this can be simply a furnace that concentrates sunlight enough to melt a charge of regolith, causing the constituents with low boiling points to vaporize, and then they can be pumped off, chilled, and processed. If done in large enough quantities, this is worthwhile just for the hydrogen, carbon, and nitrogen that exist at levels of around 50 to 100 ppm in lunar soil. A fully melted regolith mass which is spun and cooled in a centrifuge will have certain minerals separate out. This is useful for production of glass. Electrolysis done in a charge of molten regolith produces pure metals that sink and oxygen that bubbles off. An alternative that produces similar results is the Schubert dust roaster, an approach too novel to summarize. Each approach has its pros and cons.

Just as it would usually be preferable to build reactor vessels that operate at higher temperatures so they consume less reactant or use one that is more common, a furnace that uses purely sun to produce useful products is usually preferable to using any chemical reaction at all. This becomes more true as infrastructure accumulates allowing furnaces to be larger, and made more from local materials or materials that at least don't come from Earth. As the timeline progresses, the energy supply on the Moon also becomes bigger and bigger. Because most of the energy supply is derived from plentiful and reliable sunlight, the advantage of energy-based approaches increases with time. Even at the beginning of settlement, approaches that use chemical reactions can compete only where the reactants are plentiful, are nearly completely recycled, and use simpler, cheaper vessels and less energy.

An example of each of these approaches is examined following the chart.

Chemical makeup of lunar regolith.

HL = Highlands, LTB = Low-Ti Basalt, HTB = High-Ti Basalt.
From R. D. Waldron's chapter in Resources of Near-Earth Space
Things to remember about lunar chemistry Because water was not involved in the Moon's formation, and exists only at the poles, and only as captured vapors from meteor impacts, the absence of water in lunar minerals gives them somewhat different properties than Earth equivalents. One anticipated effect is that glass made of lunar materials will have much greater tensile strength due to the absence of the microscopic flaws created by water molecules in the matrix.

All lunar minerals are highly reduced, meaning they tend to have extra electrons. This also changes their chemical properties, but it won't be clear how until we can experiment on-site.
Metal
Oxide
Pyroxene [?]
(Ca,Fe,Mg)2Si2O6 [?]
Olivine
(Mg,Fe)2SiO4
Plagioclase
(Ca,Na)(Al,Si)4O8
Opaques
(mostly Ilmenite, FeTiO3)
HL [?]
(5 - 35) [?]
LTB
(42 - 60)
HTB
(42 - 60)
HL
(0 - 35)
LTB
(0 - 36)
HTB
(0 - 10)
HL
(45 - 95)
LTB
(17 - 33)
HTB
(15 - 33)
HL
(0 - 5)
LTB
(1 - 11)
HTB
(10 - 34)
SiO2 [?] 51 - 55 41 - 54 44 - 54 38 - 40 34 - 38 29 - 39 44 - 48 44 - 48 47 - 53 ≤0.1 <1 <1
Al2O3 [?] 1 - 3 1 - 12 1 - 6 ≤0.1 0 0 32 - 36 32 - 35 29 - 35 1 - 65 ≤1.2 ≤2
TiO2 0.5 - 1.3 0.2 - 3 0.7 - 6 ≤0.1 0 0 ≤0.3 0 0 0.4 - 53 51 - 54 52 - 74
Cr2O3 0.3 - 0.7 ≤1.5 ≤0.7 ≤0.1 0.3 - 0.7 0.1 - 0.2 trace 0 0 0.4 - 4 0.2 - 0.8 0.4 - 2.2
FeO 8 - 24 13 - 45 8 - 46 13 - 27 21 - 47 25 - 29 0.2 - 0.3 0.4 - 3 0.3 - 1.4 12 - 36 44 - 47 15 - 46
MnO 0 ≤0.6 ≤0.7 0 0.1 - 0.4 0.2 - 0.3 0 0 0 0 0.3 - 0.5 <1
MgO 17 - 31 0.3 - 26 2 - 23 33 - 27 19 - 39 34 - 37 ≤0.2 0.1 - 1.2 ≤0.3 7.7 - 20 0.1 - 2.3 0.7 - 8.6
CaO 2 - 17 2 - 17 4 - 21 0.2 - 0.3 ≤0.3 0 19 - 20 17 - 19 14 - 19 ≤0.6 <1 <1
Na2O 0 ≤0.1 ≤0.2 0 0 0 0.2 - 0.6 0.4 - 1.3 0.7 - 2.7 0 0 0
K2O 0 0 0 0 0 0 ≤0.15 ≤0.3 ≤0.4 0 0 0

On Earth this mineral is different. It often includes sodium (Na) or zinc(Zn), and aluminum is so typical it is part of its formula.

The formula means that 2 atoms that could be calcium, iron, or magnesium combine with 2 atoms of silicon and 6 atoms of oxygen. The crystal lattice of the rock alternates between oxygen atoms and the others.

Figures in this row are the percent of volume taken up by the mineral above in rocks and soil of each type.

For instance, between 5 and 35 percent of the volume of highland rocks and soil are made of pyroxene. This covers a wide range of different rocks made of many different minerals that fall under the general category of pyroxenes. The usefulness of all these figures is to give a bit of an idea how much you can increase the content of desired elements in the rocks you gather by looking for ones mostly composed of certain minerals. Soil of preferred composition can also be collected by color or location.

Silicon dioxide, or silica - glass is mostly this. Pure, as a glass it is fused quartz, a stong, non-reactive glass that transmits most wavelengths - including ultraviolet light (for a tan, or a sunburn). It can withstand very high temperatures. It has an extremely low coefficient of thermal expansion, meaning it shrinks and expands only a tiny amount when temperatures fall and rise. So, seals around the edges of glazing made of it only need to fill small gaps and aren't stretched much when it's cold. Tubes made of it can run through heat exchangers and radiators, without cracking. At pressure of 1 atmosphere, it softens at 1665°C. It can be drawn into fibers with high optical quality (fiberoptics).

Aluminum oxide, or alumina - in crystalline form, it is sapphire. It is very hard, and like silica very clear, transmitting light from 150 t0 5500 nm. In its amorphous form, it is a hard, white material with good compressive strength that can be used as a refractory material up to temperatures of 1700°C.

Solar Furnace

A vessel loaded with regolith and rock that is heated by the sun can be quite simple and work at a low temperature, or quite complex and work at a high temperature.

Regolith heated to 1000°C releases most of the volatile elements present in tiny quantities in it. This temperature is low enough that many materials can easily withstand it. A simple parabolic mirror could be used to concentrate sunlight, that is then bounced and refocused a couple of times to send it through the window of a vessel filled with regolith. That vessel could rotate like the drum of a cement mixer to spread the heat through the drum's contents by mixing. Thus the whole mass can be brought up to the desired temperature without significant hotspots and without the intense sun ever touching the drum's walls. The proportion of the regolith that would be released as vapors over this process would in total be only about 0.1% of its mass. This is easily pumped off during the process. It is only a kilogram of material for each metric ton, and usually more than half of that would be sulfur. The equipment needed to do this together masses enough that this would be a net gain after less than 1000 tons are processed - the chemicals acquired would mass more than the equipment needed to acquire it, which is mostly light, simple, and durable. The frames to support all the parts can be manufactured locally early in the timeline. Local manufacture of mirrors is also an early priority.

Regolith melts to a thin liquid by the time it is 1200°C. However, much of the contents of that liquid are tiny crystals, which are suspended in the liquid matrix. If you spin this lava in a centrifuge, the crystals will settle out. It shouldn't take spinning equivalent to very many gravities to make that happen in a timely manner. Then the liquid portion (which must still be near 1200°C) needs to be poured off or drawn off. The solid portion remaining will be olivine, which is composed of oxides of magnesium, iron, and silicon. The liquid will be oxides of aluminum, calcium, titanium, and sodium, mixed with the remaining leftover silicon dioxide and a small amount of other substances. Each substance will have impurities of the other, it would require experimentation to see how pure a result this technique can yield.

If the separation by centrifuge was done at a slightly lower temperature, the crystals would be pyroxene, which has a broader ranger of chemistries. The remainder in that case would be more purely oxides of aluminum, silicon, calcium, and sodium. Pyroxene has a mineral structure that inclueds magnesium, iron, and silica, like olivine does, but can also include manganese, titanium, cobalt, chromium, and others.

This sort of separation is known to happen on Earth in magma chambers (see sidebar). If this approach was experimented with to see if other types of separation can be done this way, more such techniques might be found. Though the results might not be very pure, it could still yield minerals with desirable properties. Or, the technique could be repeated one or more times to improve purity. Production of transparent glass in bulk could be made much easier by development of this technique, as discussed under Glass Production.

Solar Thermal Reactor

If heating is combined with chemical reactions there are many ways to create one desired product and purify it. The processes that have been considered for the Moon have focused on the use of hydrogen or carbon chemicals as the reactants.

Hydrogen is attractive for the processing of ilmenite (FeTiO3) to produce iron, water, and titanium dioxide. Simple pre-processing methods can be helpful - sieving to select the smallest particles, and use of magnets to draw out the ones with the most iron. This was proposed mostly to acquire oxygen by then splitting the water into hydrogen and oxygen, but since it is probably easier to get oxygen from water at the poles, it could ultimately find greater application in the production of iron. Titanium dioxide is also useful.

There are a range of carbothermal reactions (ones that use carbon chemicals and heat) to refine metals. It has been considered for production of aluminum, silicon, magnesium, calcium, cobalt, nickel, and iron. Pure powdered carbon is usually used, or sometimes methane. The products if the reaction is complete are carbon monoxide and carbon dioxide, and hydrogen and water if methane was used. The process on Earth is often complicated by the production of carbides (such as Al4C3) and oxides being left over after the reaction. What is usually done to improve yields is to do the process in a vacuum furnace at higher heat. Once again, something that is greatly aided by doing it on the Moon. The key to making carbothermal processing attractive is almost complete recovery and recycling of the gases produced, through cheap processes.

This reactor uses the heat of the sunlight concentrated in the mirrors to melt the regolith loaded into the silo. Then hydrogen is bubbled through the mix. The reaction of hydrogen with the molten mix produces water and free metals, mostly iron. This model looks at processing ilmenite this way. If the reactor's contents can be made hot enough, other metals can be purified this way. Pre-processing elsewhere can also make this unit more versatile.

Molten Regolith Electrolysis (MRE)

Molten regolith is conductive enough for electrolysis to work in it without anything more being added. As electric current passes through the melt, the stream of electrons cause molecular bonds to break, producing ions. The oxygen ions are attracted to the positive charge of the anode, and the metal ions are attracted to the negative charge of the cathode. By placing the anode at the top of the melt chamber, the oxygen can be drawn off of the top of the melt, and with the cathode at the bottom, the molten metals can be drained from the bottom.

Current designs contain the melt in a chamber that is larger than the melt pool created when current is passed through it. A crust of solidified lava is allowed to form by the walls of the chamber, reducing the amount of heat the chamber itself must withstand. When operating, the electrolysis melts only a ball of regolith at the center of the chamber. To add fresh regolith, a hole must be punched through the crust.

The device produces a metal puddle that is an alloy of all the metals present in the regolith, including several that are very difficult to purify in other ways - titanium, magnesium, aluminium. Further processing would be needed to purify them into single elements or or divide some metals from others to create useful alloys. Once the process is rolling almost all the current used leads directly to the oxygen and metal products. However, the overall efficiency of the process depends on how much energy is required to turn the highly mixed alloy into useful alloys and pure metals. If this depends on the use of reactants that must be imported from Earth, that is also a big strike against it. Possibly seiving, selection of feedstock for a favorable mineral blend, and pre-processing to remove some metals (notably iron), could reduce this issue enough that MRE outcompetes its rival, the Schubert dust roaster discussed in the next section.

One disadvantage of this system is that the only really effective substance the anode can be made of is iridium. The price of iridium has often been above $1000/oz in recent years. If the anode is any other substance, it combines with the oxygen ions in the melt and disintegrates in short order. Even an alloy of iridium with some other metal (tungsten has been tried) will have to be replaced so much sooner you might as well just use pure iridium.

As it happens, iridium is a platinum group metal and is typically present in metal asteroids at levels that would qualify as high-grade ore on Earth. Metal asteroids are also rather common. Once mining of asteroids is well underway, possibly it would be cheaper to get iridium from them, than from Earth.

Deflected Plasmas - Schubert Dust Roaster

This apparatus needs much more development than molten regolith electrolysis to attain a working prototype. It is solidly founded on a set of methods that have each been applied successfully for purposes similar to the part they play in it, but they have never been used together, and certainly not to do something so challenging. It's novel enough that probably a large set of prototypes would need to be built and tested in order to design a good working model. Because local gravity and ambient pressure have a big effect on testing this device, it's the sort of thing that might only see development once we have the facilities on the Moon to do that.

This device produces a range of pure metals, pure oxygen, and slag that makes good refractory materials. It is superior to MRE in certain respects. The metals it produces collect separately as pure elements, not as a molten puddle of mixed metals. It doesn't require rare, expensive iridium. It's unlikely to be nearly as energy efficient, though. The system is also more complex and would seem to require more maintenance, at the least to remove deposits that build up on the walls. If it performs better in durability and reliability, it could be the better option if energy is plentiful enough that the lower efficiency is not a concern. When considering efficiency, the energy needed to turn the metal alloy produced in the MRE units into pure metals or useful alloys would have to be included, which could make the two systems a lot more comparable in total energy consumption for a given product.

The designs in the linked documents are for a unit delivered from Earth that must work after landing with minimal setup or maintenance. Units manufactured on the Moon have no need to consider payload mass and can be designed to get ongoing maintenance. The top operating temperature in that case can be higher than the 2700°C upper limit in these designs. Locally made dust roasters portrayed in the colonies will be much more massive. All the structural pieces will be thick enough to account for tubing running through them, carrying coolant. They will all have liners so during maintenance, the unit can be opened up, the liners replaced, and the deposits on the old liners be chipped or ground off. If cooled properly, that allows a unit to draw more power and super-heat the plasma stream above the dissociation energy of all the component metal oxides. Every metal in the regolith could be refined out, even calcium and magnesium. Since the unit is designed to act on the molten stream and condensed globules that never touch the device's walls, the proper modicum of cooling won't affect output.

The model notes go through each stage of processing. The portrayed unit is based on this AIAA paper. It is estimated to draw 200 kW of power to produce 17 metric tons of silicon per year, plus a quantity of aluminum and iron. Units made on the Moon would have different dimensions. In particular, the structural parts would all be thicker and heavier, and include systems to actively cool them.

Glass Production

Clear glass can have a wide range of chemistries. The glass used in windows on Earth is soda-lime glass, and is composed mostly of silicon dioxide, with some sodium oxide, and a small amount of a few other things. Pure silica sand, composed almost entirely of silicon dioxide, is common on Earth. The Moon has nothing like that. The proportion of silica in the minerals of the regolith is never more than half.

Most of the metal oxides in the regolith will form clear glass, if melted and then properly cooled to form a solid without a crystal structure, which is technically what glass is. The oxides of aluminum, magnesiium, calcium, and sodium will all do this just like silica does. Those metal oxides form 80 to 90 percent of the volume of most lunar minerals. Working with glass that is high in metal oxides other than silica can be more difficult - it could require higher heating and the glass may only be workable over a narrower temperature range.

A solar glass furnace would have little problem attaining the higher temperatures needed. The furnace and all glass forming equipment need to be made of materials that can withstand about 1200°C - no higher than all the equipment used to produce the refined feedstock for making the glass, and much less than the metal refining techniques discussed above.

The vacuum makes it easier to work with glass that has a narrow working temperature range. Heat is lost slowly and that can be slowed much more, simply by surrounding the working area in foil that reflects heat back into the system. Rather than having to reheat glass to anneal it (to relieve internal stresses that weaken the glass), the glass must be cooled, enough to have the desired viscosity to be put through rollers that extrude it as flat sheets. Sheet glass that is flat, or even better that has a curve to it, is what is really needed in giant quantities for many of the structures in the colonies. Radiators that gently chill the rollers can be used to put the glass at the right temperature to be properly formed, after which just putting a reflective tarp over the hot glass will slow cooling enough for it to anneal properly.

However, the 10 to 20 percent of the regolith mineral mix that isn't oxides of silicon, aluminum, calcium, magnesium, and sodium, must be removed almost completely or the glass will be strongly tinted or opaque. Iron turns it blue-green, titanium and sulfur yellow-brown, manganese purple, chromium dark green, and all of these are common enough in the regolith to have a strong effect. And there are a long list of others, some of which will discolor strongly at very low concentrations. Cobalt will yield blue glass at concentrations of 250 ppm.

Finding a deposit of highlands regolith that is very low in the undesired metals is the first thing to do. This shouldn't be too hard, and results could be much better if material can be obtained from rock instead of fine-grained regolith. Some highlands regolith is almost entirely plagioclase, so possibly the bedrock it came from is pure plagioclase. Plagioclase has none of the metals that are problematic, it's composed purely of oxides of light metals - aluminum, calcium, sodium, and silicon. If particles containing iron can be removed with magnets from powder plagioclase regolith, the remainder might be pure enough to make decently clear glass. The titanium is all in ilmenite, which also contains iron so would be removed by the magnetic separation. That might be enough.

Something of a tint to windows is vastly preferable to not having windows, and the lunar landscape is so colorless a tint to it could be regarded as an improvement. If glass could be produced in industrial quantities with the above process, which is fairly low-energy and low-tech, it would be an attractive option early on even if the resulting glass had a pretty clear tinge to it. But there is also another option.

The process of purifying olivine and pyroxene mentioned above in the solar furnace section could be applied to making clear glass. All of the minerals most common in the regolith that cause tinting fit into the crystal structure of pyroxene. The liquid poured off after pyroxene crystals have been pressed to the bottom of a vessel in a centrifuge will be made almost completely of oxides of silicon, aluminum, calcium, and sodium. Remaining minor components of the liquid aren't problematic - oxides of phosphorus and potassium, which are used in making glass on Earth and don't affect transparency.

As long as the regolith was toasted for a good while to remove its volatiles before being melted, this process as well could provide decently clear glass. The toasting especially needs to remove the sulfur, which otherwise will color glass yellow. This process is also technically simple. It requires more energy to perform, but mostly as heat which can be obtained from the sun, essentially for free once the equipment exists.

If purification by centrifuge worked well enough, any regolith feedstock could be used for this process. If it didn't work well enough the first time, repeating it could result in the purity required. Or, once again, selecting regolith from the highlands with a low content of troublesome minerals to begin with could yield a product good for clear glass after the first time through the centrifuge. Or, a centrifuge that presses the molten regolith with greater force could increase purity sufficiently. After all, centrifuges are another thing the Moon is very kind to - no atmospheric drag on the spinning vessels, and making frictionless magnetic bearings is a lot easier in the low gravity. Centrifuges benefit from this the same way flywheels do.

Freedom to include large quantities of clear glass in lunar architecture is tremendously beneficial. Where glazing uses the radiation blind design, or windows are sunk at the bottom of shafts that eliminate direct sight-lines to the sky, glass can be just a few centimeters thick - say, 8 cm (3"). That is still pretty thick, because it needs to support the pressure of the atmosphere, and that is a lot of force. The glass would need to be made of several laminated panes, to improve strength and prevent failure by shattering. If it can be thicker, then panes can be larger. Glazing means natural sunlight and a view of the sky. If you think it wouldn't bother you to live without either of those things, try it for a few days. A colony must have this, and have it over a sizable area so that everyone can enjoy it as much as they want.

Both of these processes can be scaled up as much as needed. With the infrastructure assembled to build a large number of giant glass furnaces, plate glass rolling equipment, and associated paraphernalia, Cernan's Promise is able to undertake construction of Lalande City. Such a city would not be possible otherwise.

Excavation

The powder and fine-grained regolith on the surface that can simply be scooped up is good enough for many things, with the designs adapted for that shown in previous sections, including the MIP stations, STeMP units, solar furnaces, and MRE units.

For purer minerals, at Lalande Crater there is a wealth of rocks exposed on the surface, with a variety of compositions. Some are chunks of the bedrock that was blasted sky high during the impact that created Lalande, and some of that bedrock may well have a composition that is advantageous for certain applications. For instance, some of it might be pure plagioclase, as mentioned in Glass Production. Hauling the good rocks even 20 or 30 kilometers could well be worth it. What kinds there are around Lalande Crater will need to be mapped and catalogues.

The north pole doesn't have many rocks, as it is a far older part of the Moon. In such places fine-grain regolith has long since buried larger rock chunks, leaving them many meters under the surface. That's okay - there are other reasons why it's good to make big, deep, steep-walled holes on the Moon than simply mining. That's where you are going to put your habitats.

Under the first 20 cm or so of powdery regolith (8 inches), the ground quickly becomes very hard packed. The individual particles of the soil look like shrapnel, because that is basically what they are - blast debris. Some of them are glass globules originating from ground close enough to an impacting asteroid it was all instantly melted. The molten spray of droplets cooled solid before they hit the ground. The rest is shrapnel. That seems to be why the stuff clumps so well, the jagged, oddly shaped grains sort of gnarl together. So breaking it up isn't easy.

Provisionally, the excavation for the habitats is planned to be done by blasting, just a teeny bit. The kind of blasting done in quarries on Earth wouldn't be safe unless all your machinery was moved quite a distance away. It would leave a lot of dust and debris to be cleaned up too. Since dust on moving parts is a serious problem on the Moon, you want to avoid anything that might lead to that. Instead, the idea is to bore holes into the ground, by running fiberoptic cables down the center of a tube, which carry concentrated sunlight. Then micro-charges of explosive are placed. When detonated, they are just enough to loosen the soil. They don't throw anything into the non-existent air.

It would take experimentation to find the depth and explosive energy needed for the right result. To give the soil somewhere to expand into, a trench needs to be dug around the grid of boreholes. That way the energy of the blasts can be released as movement to the the sides as well as movement upwards and compaction, resulting in loose soil. Boring by melting the ground with the sun is important. A conventional drill would heat up so fast in the vacuum and high-friction soil it would seize in no time. If that was overcome, there is an excellent chance dust would get into the moving parts and break them.

Once the soil is broken up, a gantry crane like the one shown in the Construction section is used to clear the debris. To go deeper, a new trench is dug (or rather, mostly melted) and the process is repeated. If you reach bedrock, hopefully it will be composed of a useful mineral, or a strata not far down will be.

Boring and trenching becomes a different challenge at that point. Heat diffuses through solid rock many times faster than through the regolith. The sun gun won't be very effective any more. Quite possibly the thing to do then would be to clear the regolith from a large section of bedrock, cover the area well enough to fill the space with a very thin atmosphere, and switch to conventional drills and saws.

When you need to do conventional digging, machinery will need to borrow mass from elsewhere by either being anchored to a cable set in something suitably immovable, or being loaded with ballast. Otherwise, in the low gravity the force needed to dig would be more than enough to lift up the whole digger, which will look silly for a moment and then tip over. Robotic diggers will need to sense their balance well enough to brace themselves, move ballast on an arm to compensate forces correctly, and know when and where to use anchors and cables. As long as they can, they can use conventional rock ripper buckets to dig even through solid rock.[?]

three pronged ripper attachment for an excavator, showing two sharp points on the edge of each prong
Courtesy digrock.com